Abstract: This study investigated a typical small watershed surrounding a stone coal mining area in southern China. Based on concentrations of eight heavy metals (Cd, Pb, As, Hg, Cr, Cu, Zn, and Ni) from 1,823 agricultural soil sampling sites, the geo-accumulation index, pollution index, potential ecological risk index, and positive matrix factorization (PMF) model were integrated to systematically characterize heavy metal pollution, quantify source contributions, evaluate ecological risks, and establish a quantitative linkage between pollution sources and ecological risk contributions. The results showed that Cd was the most prominent contaminant in the study area. The mean geo-accumulation index and pollution index of Cd were 1.85 and 1.59, respectively, indicating moderate accumulation and moderate pollution levels, with 45% of sampling sites exceeding the risk screening value. Hg exhibited a relatively high ecological risk, with a mean potential ecological risk index of 80.55, whereas the remaining elements showed relatively low pollution levels. The PMF model identified four major pollution sources: composite sources of geological background and agricultural activities (48.20%), mixed sources of fuel combustion and transportation (35.87%), industrial sources associated with mining, beneficiation, and smelting activities (7.83%), and mixed sources of mining and smelting industries and livestock farming (8.11%). The mean comprehensive potential ecological risk index (RI) was 353.03, indicating a considerable ecological risk level, with Cd and Hg identified as the dominant risk contributors. Further analysis revealed a significant discrepancy between the mass loading contribution and ecological risk contribution of different pollution sources. Although the industrial source associated with mining and smelting activities contributed only 7.83% of the heavy metal mass loading, it accounted for 57.67% of the total ecological risk, showing a typical pattern of “low mass contribution but high risk contribution”. Therefore, this source should be prioritized for risk management and control. The source-risk linkage framework established in this study provides a scientific basis for precise source identification, risk assessment, and zonal management of heavy metal pollution in farmland soils surrounding stone coal mining areas.